284 JOURNAL OF LAW AND POLICY
improve the reliability of their conclusions. In my own essay in
this volume, I explore the advantages and disadvantages of the
various techniques employed in this volume, commenting on
how the legal system tends to react to them. While it might be
too ambitious to predict that the various methods will converge,
there is already reason for optimism that the insights of the
various approaches will influence those of the others, creating a
field with a healthy combination of cooperation and competition.
Thus, in this volume, we see the expanded use of linguistic
features by the computational researchers, while at the same
time we see efforts by stylistic researchers to introduce statistical
modeling into stylistic authorship analysis.
A quick glance at the table of contents shows a lot of
contributors with Ph.D.s and very few with law degrees, an
unusual array for a law journal publication. Nonetheless, the
legal academic community is very much present in this volume.
In addition to my own essay, comments by two prominent law
professors who specialize in scientific evidence—Edward Cheng
and Jonathan Koehler—consider the legal community’s likely
response to the advances in authorship attribution described by
the linguists.^2 Moreover, participating in the workshop were two
statistical “consultants” (Stephen Fienberg of Carnegie Mellon
University and Robert Carpenter of Columbia University).
This interaction between the scientific community and the
evidence scholars was one of the workshop’s main goals.
Forensic identification sciences have been under severe attack as
inadequately grounded in science over the past decade—largely
for good reason. Just as the scientists had a lot to learn from the
reactions of the legal scholars to their work, we believed that the
legal scholars could benefit from seeing in action a relatively
young forensic science that takes itself seriously as science.
While the evidence literature decries the absence of concern
over the rate of error in one forensic science after another, an
annual workshop on authorship and plagiarism identification
actually requires that algorithms presented be subjected to a
(^2) Also present at the workshop were D. Michael Risinger and Michael
Saks, two additional prominent scientific evidence scholars.